The function and regulation of the U11-48K protein in U12-dependent splicing

Transcription

1 The function and regulation of the U11-48K protein in U12-dependent splicing Janne Turunen Institute of Biotechnology and Division of Biochemistry and Biotechnology Department of Biosciences Faculty of Biological and Environmental Sciences and Helsinki Graduate Program in Biotechnology and Molecular Biology University of Helsinki ACADEMIC DISSERTATION To be presented for public criticism, with the permission of the Faculty of Biological and Environmental Sciences of the University of Helsinki, in auditorium 2 in Information Centre Korona (Viikinkaari 11, Helsinki) on January 20 th 2012, at 12 noon. HELSINKI 2012

2 Supervisor Reviewers Docent Mikko Frilander Institute of Biotechnology University of Helsinki Helsinki, Finland Docent Minna Poranen Department of Biosciences and Institute of Biotechnology University of Helsinki Helsinki, Finland Professor Anders Virtanen Department of Medical Biochemistry and Microbiology Uppsala University Uppsala, Sweden Opponent Custos Professor Mark McNally Department of Microbiology and Molecular Genetics Medical College of Wisconsin Milwaukee, Wisconsin Professor Kari Keinänen Department of Biosciences University of Helsinki Helsinki, Finland ISBN (paperback) ISBN (PDF) ISSN Cover layout by Anita Tienhaara Helsinki University Print, Helsinki 2012 ii

3 "Every model we make tells us how our mind works, as much as it tells us about the universe" Robert Anton Wilson iii

4 Table of contents List of original publications... vi Abbreviations... vii Summary... viii 1 Review of the literature Introduction Overview of eukaryotic pre-mrna processing Co-transcriptional nature of pre-mrna processing Variation and origins of spliceosomal introns Prevalence of introns Introns and evolution Relationship of spliceosomal introns to self-splicing introns Minor and major introns Intron structure Characteristics of U2-type and U12-type introns Conversion of intron types, and evolution of U12-type introns Significance of U12-type introns Spliceosome components Components of the U2-dependent spliceosome Components of the U12-dependent spliceosome Spliceosome assembly and catalysis Splicing catalysis Assembly of the U2-dependent spliceosome Is the spliceosome a ribozyme? Assembly of the U12-dependent spliceosome Comparison of initial intron recognition in the two spliceosomes Splice site definition and alternative splicing Initial splice site definition over exons Splicing enhancers and silencers Alternative splicing increases proteome complexity Trans-acting factors affecting splice site choice SR proteins HnRNP proteins G-run motifs and the hnrnpf/h protein family Core spliceosomal components as splicing regulators Complex splicing regulatory elements Other factors affecting alternative splicing Alternative splicing of U12-type introns Quality control and nonsense-mediated decay Nuclear degradation of aberrant transcripts Splicing is linked to mrnp export Nonsense-mediated decay Conserved sequence elements and regulation of splicing factors through nonsense-mediated decay Most NMD events are caused by splicing errors Conserved NMD-inducing elements in splicing factor genes Feedback regulation of splicing factors iv

5 2 Aims of the study Materials and methods Results and discussion U11-48K protein recognizes the U12-type 5' splice site and is essential for U12-dependent splicing (I) U11-48K specifically recognizes the U12-type 5' splice site during initial intron recognition Parallels between 5' splice site recognition and cross-intron interactions in the two spliceosomes U11-48K contributes to U11/U12 di-snrnp formation or stability The level of U11/U12 di-snrnps is regulated through a conserved feedback mechanism (II) Highly conserved sequence elements in the genes encoding U11-48K and U11/U12-65K The USSE directs alternative splicing and mrna destabilization Activation of AS-NMD in the U11-48K transcript is regulated by multiple factors (II, III) The USSE binds U11/U12 di-snrnps (II) AS-NMD is suppressed by hnrnpf/h and U1 snrnp (II, III) Evolutionary implications of the USSE Concluding remarks Acknowledgements References v

6 List of original publications The thesis is based on the following articles, which are referred to in the text by their Roman numerals: I II III Turunen, J.J.*, Will, C.L.*, Grote, M., Lührmann, R., and Frilander, M.J. (2008). The U11-48K protein contacts the 5' splice site of U12-type introns and the U11-59K protein. Molecular and Cellular Biology 28: (Reproduced here with permission from American Society for Microbiology) Verbeeren, J.*, Niemelä, E.H.*, Turunen, J.J.*, Will, C.L., Ravantti, J.J., Lührmann, R., and Frilander, M.J. (2010). An ancient mechanism for splicing control: U11 snrnp as an activator of alternative splicing. Molecular Cell 37: (Reproduced here with permission from Elsevier) Turunen, J.J., Verma, B., Frilander, M.J. HnRNP F/H, U1 snrnp and U11 snrnp co-operate to regulate the stability of the U11-48K pre-mrna. Unpublished manuscript. * Equal contribution The author's contribution to each publication: I II III JJT participated in planning the experiments and performed the in vitro characterization of the protein-rna interactions as well as the analysis of the effect of U11-48K RNAi knockdown on in vivo splicing, and wrote the article together with the other authors. JJT planned and performed the in vitro characterization of the factors involved in recognizing the regulatory elements in both the U11-48K and U11/U12-65K transcripts, and wrote the article together with the other authors. JJT planned and performed the majority of the experiments, and wrote the manuscript. vi

7 Abbreviations 3'ss 3' splice site 3' UTR 3' untranslated region 5'ss 5' splice site aa amino acid(s) AS-NMD alternative splicing coupled to NMD bp base pair(s) BP branch point BPS branch point sequence CERES composite exonic regulatory elements of splicing CTD C-terminal domain (of RNAPII) EJC exon junction complex ESE exonic splicing enhancer ESS exonic splicing silencer hnrnp heterogeneous nuclear RNP ISE intronic splicing enhancer ISS intronic splicing silencer LECA last eukaryotic common ancestor mrna messenger RNA mrnp messenger RNP NMD nonsense-mediated decay NMR nuclear magnetic resonance nt nucleotide(s) PPT polypyrimidine tract pre-mrna precursor mrna PTC premature termination codon qrrm quasi-rrm R purine RNAi RNA interference RNAPII RNA polymerase II RNP ribonucleoprotein RRM RNA recognition motif RS arginine-serine-rich (domain) RSV Rous sarcoma virus snorna small nucleolar RNA snrna small nuclear RNA snrnp small nuclear RNP SR serine-arginine-rich (protein) SRE splicing regulatory element TEC transcription elongation complex TREX transcription-export complex USSE U11 snrnp-binding splicing enhancer Y pyrimidine ZnF zinc finger vii

8 Summary The removal of noncoding sequences, or introns, from the eukaryotic messenger RNA precursors is catalyzed by a ribonucleoprotein complex known as the spliceosome. In most eukaryotes, two distinct classes of introns exist, each removed by a specific type of spliceosome. The major, U2-type introns account for over 99 % of all introns, and are almost ubiquitous. The minor, U12-type introns are found in most but not all eukaryotes, and reside in conserved locations in a specific set of genes. Due to their slow excision rates, the U12-type introns are expected to be involved in the regulation of the genes containing them by inhibiting the maturation of the messenger RNAs. However, little information is currently available on how the activity of the U12-dependent spliceosome itself is regulated. The levels of many known splicing factors are regulated through unproductive alternative splicing events, which lead to inclusion of premature STOP codons, targeting the transcripts for destruction by the nonsense-mediated decay pathway. These alternative splice sites are typically found in highly conserved sequence elements, which also contain binding sites for factors regulating the activation of the splice sites. Often, the activation is achieved by binding of products of the gene in question, resulting in negative feedback loops. In this study, I show that U11-48K, a protein factor specific to the minor spliceosome, specifically recognizes the U12-type 5' splice site sequence, and is essential for proper function of the minor spliceosome. Furthermore, the expression of U11-48K is regulated through a feedback mechanism, which functions through conserved sequence elements that activate alternative splicing and nonsense-mediated decay. This mechanism is conserved from plants to animals, highlighting both the importance and early origin of this mechanism in regulating splicing factors. I also show that the feedback regulation of U11-48K is counteracted by a component of the major spliceosome, the U1 small nuclear ribonucleoprotein particle, as well as members of the hnrnp F/H protein family. These results thus suggest that the feedback mechanism is finely tuned by multiple factors to achieve precise control of the activity of the U12-dependent spliceosome. viii

9 Review of the literature 1 Review of the literature 1.1 Introduction Splicing of messenger RNA precursors (pre-mrnas) is a ubiquitous eukaryotic process that affects virtually all protein-coding transcripts in humans, and thus has wide implications for normal gene expression, as well as for disease (reviewed by Wang and Cooper, 2007). However, the discovery that eukaryotic genes are discontinuous (Berget et al., 1977; Chow et al., 1977) was initially met with amazement and even incredulity. Understandably, the reason for maintaining such apparently useless DNA was initially difficult to fathom. Even more puzzling was the question of how the intervening regions, or introns, as they became known, were properly recognized and removed from the expressed regions, or exons. There seemed to be very little information in the introns themselves, and no pattern similar to the well-known protein-encoding genetic code was to be found. Yet, the precise removal of introns was crucial to maintaining the correct reading frame in the resulting messenger-rna (mrna), and thus for the expression of the correct proteins. The observation that some of the small nuclear RNAs (snrnas) bore sequences compatible to the splice site sequences (Lerner et al., 1980; Rogers and Wall, 1980), paved way for the realization that the splicing machinery, the spliceosome, is itself a ribonucleoprotein (RNP) complex. While highly degenerate, the splice sites are nonetheless recognized by snrna-containing small nuclear RNPs (snrnps) through base-pairing interactions with the snrnas (Mount et al., 1983; Black et al., 1985). The situation became even more complex with the discovery of alternative splicing. This process was originally found in immunoglobulin transcripts, the alternative isoforms of which produce the membrane-bound and soluble antibodies (Alt et al., 1980; Early et al., 1980; Rogers et al., 1980). While this observation offered an insight into how splicing could be useful, i.e. by producing mrnas for alternative protein isoforms from one gene (reviewed by Breitbart et al., 1987), it underscored the inherent puzzle in splicing: How are the proper splice sites selected? It became clear that the splice site sequences themselves were not the only determinants of splicing, as the recognition of splice site sequences was found to be affected by exonic sequences (Somasekhar and Mertz, 1985; Reed and Maniatis, 1986). Over time, it was recognized that splicing regulatory elements play a crucial role in splice site activation in both constitutive and alternative splicing. Alternative splicing was also found to be a more common event than originally expected, particularly after completion of the human genome sequencing project (Lander et al., 2001), when it became obvious that the complexity of the human proteome heavily depends on alternative splicing to produce the protein isoforms that are many-fold more numerous than the genes encoding them (reviewed by Nilsen and Graveley, 2010). Almost as surprising as the discovery of splicing itself was the discovery of a rare class of introns (Jackson, 1991; Hall and Padgett, 1994) that were found to be excised by a separate spliceosome with its own set of snrnps analogous to, but distinct from those of the canonical major spliceosome (reviewed by Patel and Steitz, 2003). Why should this be? What possible 1

10 Review of the literature use could an organism have for two machineries performing essentially identical tasks? The low frequency, inefficient excision and extremely conserved splice site sequences of these minor, or U12-type introns, as well as their absence from many organisms, initially suggested that they may be molecular fossils, slowly being purged from the genomes of eukaryotes. However, their retention in homologous locations over hundreds of millions of years of evolution suggests that they have important functions. The importance of U12-dependent splicing is likely linked to its regulatory role in controlling the expression of the specific set of genes containing U12-type introns and, ultimately, the downstream effects of those genes. The emerging view is that the various splicing processes are linked to each other as well as other steps of the gene expression pathway, with cross-regulation apparent at every level. 1.2 Overview of eukaryotic pre-mrna processing The initial step in the expression of proteins in all cellular organisms is transcription, during which the nucleotide sequence of DNA is copied into a complementary sequence in RNA. In eukaryotes, chromosomes are tightly packed within the nucleus of the cell, and transcriptionally active and inactive parts of the chromosomes are sequestered in different domains of the nucleus (reviewed by Lanctôt et al., 2007). To become accessible to RNA polymerase, the DNA must be unpacked from the chromatin complexes. A number of enzymes regulate the affinity of histones for DNA within the chromatin by modifying a specific set of histone amino acid residues. Once the DNA has become accessible, a set of transcription initiation factors is assembled within the promoter region of a gene, and they in turn recruit the RNA polymerase (reviewed by Valen and Sandelin, 2011). This process is regulated by enhancer and silencer elements, which can be located thousands of nucleotides away from the promoter, and exert their effect by binding regulatory factors which either promote or suppress transcription initiation. As transcription proceeds, the nascent transcript is processed in various ways, including capping at its 5' end, splicing to remove the non-coding intronic sequences and finally polyadenylation at its 3' end (see the following chapter). All of these steps are generally required before the mature mrna is transported into the cytoplasm (see chapter 1.7.2), where it can function as a template for protein synthesis. Among these steps, splicing has emerged as a highly versatile process, which not only removes introns with high fidelity, but can also give rise to alternative mrna isoforms, and thus proteins, as well as regulate gene expression by producing non-viable mrna variants Co-transcriptional nature of pre-mrna processing In eukaryotes, pre-mrnas are transcribed by RNA polymerase II (RNAPII). Together with the chromatin template, the growing transcript and other associated factors, the polymerase forms a transcription elongation complex (TEC; reviewed by Perales and Bentley, 2009). The TEC is a multifunctional assembly, which co-ordinates a large number of interdependent RNA as well as chromatin processing events (reviewed by Phatnani and Greenleaf, 2006). One of the central elements in this complex is the C-terminal domain (CTD) of the large subunit of RNAPII, composed of a large number (52 in mammals) of YSPTSPS heptapeptide repeats. The phosphorylation state of these repeats controls the initiation and elongation phases of transcription, and also co-ordinates the assembly of pre-mrna processing factors (reviewed 2

11 Review of the literature by Phatnani and Greenleaf, 2006). For example, as the transcript is initiated, the CTD is phosphorylated on the Ser5 residues of its heptapeptide repeats, allowing the loading and activation of the capping enzyme (Ho and Shuman, 1999). Towards the end of the transcript the CTD becomes phosphorylated predominantly at Ser2, and recruits and activates 3' endprocessing factors (reviewed by Perales and Bentley, 2009). The interactions are also reciprocal, with pre-mrna processing factors activating or inhibiting transcription elongation (Perales and Bentley, 2009). Whether splicing factors associate physically with RNAPII is not equally well established, although a number of interactions between them have been observed. Splicing factors are found in complexes with RNAPII (Das et al., 2007; Sapra et al., 2009), and can also activate recruitment of transcription factors to the promoter and enhance elongation (Damgaard et al., 2008; Lin et al., 2008). On the other hand, splicing factors appear to be recruited only to intron-containing, but not intronless transcripts, suggesting that RNAPII does not directly promote their binding or at least is not the only factor doing so (Listerman et al., 2006; Moore et al., 2006). Despite the lack of information on the physical relationship between splicing factors and RNAPII, the co-transcriptional nature of splicing is well documented. Looped RNAs attached to chromatin have been observed by electron microscopy (Beyer and Osheim, 1991), spliced mrnas are associated with isolated chromatin (Bauren and Wieslander, 1994; Pandya-Jones and Black, 2009) and spliced mrnas can be detected on their chromosomal gene loci by RNA in situ hybridization (Zhang et al., 1994). Also, synthetic pre-mrnas are spliced less efficiently than co-transcriptionally spliced transcripts both in vitro and in vivo, and CTD phosphorylation is required for efficient co-transcriptional splicing (Bird et al., 2004; Das et al., 2006). Finally, with long genes, the removal of most introns is completed before the pre-mrna is fully transcribed (Singh and Padgett, 2009), verifying that splicing takes place co-transcriptionally. Splicing is also linked to other steps of pre-mrna processing, including capping (Görnemann et al., 2005) and export, as well as quality control mechanisms in the nucleus, such that only correctly and rapidly processed mrnas are packages into mature messenger RNPs (mrnps) that are exported into the cytoplasm for translation (see chapter 1.7) Variation and origins of spliceosomal introns Prevalence of introns Introns are one of the features typical of eukaryotic genomes, but different eukaryotes vary widely with respect to the length and density of their introns. Some unicellular eukaryotes contain only a few introns in total in their genomes (Nixon et al., 2002; Vanacova et al., 2005), and at least one cryptophyte has lost its introns altogether (Lane et al., 2007). In contrast, multicellular eukaryotes with larger genomes typically have a larger number of introns, which are also longer and in some cases constitute a much larger fragment of their genomes than the protein-coding exonic sequences (Deutsch and Long, 1999). In humans, the lengths of the introns vary from less than 100 basepairs (bp) to several hundred kilobases (kb), with the mean length of 3300 bp (Lander et al., 2001) and the average number of introns per 3

12 Review of the literature gene (intron density) of 6.9 (Csuros et al., 2011). The most extreme case of intron density and coverage in the human genome is the gene which codes for the muscle-specific protein dystrophin. The total length of the gene is 2.5 million bp, and it contains 79 exons, which constitute only 1% of the total length of the gene (Pozzoli et al., 2002). On the other hand, in baker's yeast Saccharomyces cerevisae, a commonly used model organism in biology, only a subset of genes harbor introns, and typically no more than one intron per gene, with an average intron density of 0.1 (Csuros et al., 2011). Introns in the yeast genome are also considerably shorter than in mammals, with lengths varying between bp Introns and evolution Such a wide range of varying intron sizes and densities suggest that the importance of introns and splicing has varied in different evolutionary lineages. How, then, did this variation arise during evolution? A number of theories accounting for the origins of introns have been proposed, and they can be roughly divided into two models known as "introns-late" and "introns-early" (Darnell, 1978; Doolittle, 1978). According to the introns-late model, introns were introduced (possibly through horizontal gene transfer) to a eukaryotic ancestral proteincoding DNA genome, or independently into several eukaryotic lineages. The introns-early model postulates that introns arose before or together with protein coding sequences or else were inserted into protein-coding RNA genomes before DNA was adopted as the hereditary material. In a more stringent version of the introns-early model, known as the "introns-first" model (Jeffares et al., 1998; Penny et al., 2009), introns are seen as remnants of the original RNA world (Gilbert, 1986), representing genes for ribozymes and other functional RNAs, which later became interspersed with spacers that evolved into protein coding sequences of today. Both the introns-early and introns-first models necessitate that spliceosomal introns were already extant before the last eukaryotic common ancestor (LECA), and may also have been present in the ancestors of modern archaea and/or bacteria. Yeast is one of the most common model organisms used in splicing studies. As mentioned, the yeast genome is very intron-poor. Yeast introns typically have very precisely defined splice site sequences and are not alternatively spliced. Such a minimal, stringent splicing system has traditionally been considered to be the original primitive mode of splicing in LECA, thus lending support for the introns-late theory (Logsdon, 1998). However, recent comparisons across eukaryotes have clearly shown that the common eukaryotic ancestor instead had introns and spliceosomes very similar to most modern eukaryotes, i.e. high intron density, weakly conserved splice sites, both U2 and U12-type introns and complex spliceosomes (reviewed by Koonin, 2009; Roy and Irimia, 2009). It appears that intron loss is a common phenomenon in many eukaryotic lineages, while massive expansion in intron numbers has only taken place a few times during evolution, most notably in the common ancestor of Metazoa (Csuros et al., 2011). In very intron-sparse lineages, such as yeast, intron loss is correlated with reduction in the genome size and with the emergence of highly regular splice sites, suggesting that all these characteristics are a product of high selection pressure on these fast-reproducing organisms (reviewed by Koonin, 2009; Penny et al., 2009). Thus, introns appear to have arisen well before the LECA, but whether their origins are in the RNP or RNA world remains unresolved. 4

13 Review of the literature Relationship of spliceosomal introns to self-splicing introns In addition to spliceosomal introns, many types of organisms carry self-splicing introns, which are divided into group I and group II introns based on their different catalytic mechanisms (reviewed by Bonen and Vogel, 2001; Haugen et al., 2005; Fedorova and Zingler, 2007). They are rarely found in archaea, but are common in bacteria and in eukaryotic mitochondria and chloroplasts, and are also present in some nuclear genes encoding ribosomal RNAs. Selfsplicing introns of the same type show a high degree of structural conservation, which is required for their autocatalytic excision. They also often carry open reading frames encoding proteins that enhance the splicing activity and also enable the insertion of the excised intron into DNA, allowing the introns to function as retrotransposons. Excision of group II introns and spliceosomal introns proceeds through a similar catalytic mechanism, which has prompted the hypothesis that group II introns may be evolutionarily related to both the spliceosome and spliceosomal introns (Hickey, 1992; Stoltzfus, 1999). A number of functional similarities support this hypothesis: Both intron types are excised through a similar two-step transesterification reaction, and the structures of spliceosomal snrnas are similar to the catalytically active domains of group II introns and they even have interchangeable activities in certain cases (reviewed by Valadkhan, 2010). Furthermore, constructs based on spliceosomal snrnas can perform splicing reactions in vitro in the absence of proteins, similar to group II introns (Valadkhan et al., 2009). However, there currently is no clear evidence to show whether the spliceosome is derived from group II introns or shares a common ancestor with it, and it has been noted that group II introns and spliceosomal introns may merely have gained similar characteristic through parallel evolution due to performing similar chemical functions (Weiner, 1993). 1.3 Minor and major introns Intron structure The excision of self-splicing introns relies on specific structures within the introns themselves, and the introns are therefore highly conserved sequence elements. In contrast, spliceosomal introns are removed by a trans-acting spliceosome, and have therefore diverged significantly, with little sequence conservation between different introns. However, they do contain three regions of moderate conservation. The scissile phosphodiester bond at the 5' splice site (5'ss) is located in a conserved element, known as the 5'ss sequence, which generally includes the first six nucleotides of the intron, as well as up to three nucleotides in the exon (Figure 1). Similarly, the 3' splice site (3'ss) is preceded by a conserved intronic region. Finally, the catalytically active branch point (BP) adenosine resides within the branch point sequence (BPS) located at a variable distance upstream of the 3'ss. These are the sequences that are recognized by the core spliceosomal components during spliceosome assembly and are also involved in the catalytic steps required for intron removal and exon ligation. 5

14 Review of the literature Figure 1. Intron structure and consensus sequences. A) Schematic depiction of (U2-type) intron structure in the coding strand of a typical eukaryotic gene. Exons are depicted as boxes, and the intron as a line, with key sequence elements highlighted by letters. The locations of the scissile phosphodiester bonds at the splice sites (5'ss and 3'ss) are indicated by arrows, as well as the location of the branch point (BP) adenosine. The approximate locations of the conserved sequence elements are indicated below the intron, including the 5'ss sequence, the branch point sequence (BPS), and the polypyrimide tract (PPT) together with the 3'ss sequence. B) Human U2 and U12-type intron consensus sequences The consensus sequences of the 5'ss, BPS, and 3'ss (with PPT included) of U2 and U12-type introns are depicted schematically, with the height of the letters proportional to the frequency of the corresponding base at that position. The 5'ss, BP and 3'ss are indicated by arrows. Adapted with modifications from Roy and Irimia (2009), with permission from Elsevier Characteristics of U2-type and U12-type introns There are two parallel sets of spliceosomal introns in eukaryotes, the U2-type introns and U12-type introns (reviewed by Patel and Steitz, 2003). U2-type introns account for the majority of introns in any given eukaryote (over 99%), while U12-type introns are only found in a handful of genes, and the intron types are thus also known as the major and minor introns, respectively. The length of both intron types varies greatly, and they have similar average length (Levine and Durbin, 2001). However, there is a large subset of small U2-type introns of less than 100 bp, while such short U12-type introns are relatively rare (Levine and Durbin, 2001). This has led to the assumption that intron recognition may differ for the two intron types, with U12-type splicing relying more strictly on the so called exon definition interactions (see chapter 1.6.1). The organization of both types of introns is similar, but they differ significantly in the precise sequence composition of their splice sites and BPS, such that they can be easily differentiated from one another based on these sequences (Figure 1B). The overall consensus for the U2- type 5'ss sequence is AG/GTAAGT, with the slash denoting the 5'ss (Sheth et al., 2006). The level of sequence conservation varies greatly between different organisms, and organisms with low intron density typically also have very conserved 5'ss sequences (chapter ). For 6

15 Review of the literature example, 84 % of the introns in Cryptosporidium parvum contain the precise consensus sequence, while in human introns the corresponding figure is only 14 % (Irimia et al., 2007). The U2-type 3'ss is marked by a much shorter signal, CAG, at the very end of the intron. The U2-type BPS is generally located bp upstream of the 3'ss, and shows considerable sequence variation, similar to the 5'ss sequence (Gao et al., 2008; Corvelo et al., 2010). Between the BPS and the 3'ss lies the polypyrimidine tract (PPT), the length of which varies significantly and is generally indicative of the strength of the 3'ss, and can also compensate for weak BPSs (Corvelo et al., 2010). In contrast to the U2-type splicing signals, the U12-type introns show a much higher degree of conservation in theirs. The U12-type 5'ss sequence almost invariably conforms to the consensus /RTATCCTTT (in which R is a purine) in all organisms, although occasional deviations occur (Figure 1B; Sheth et al., 2006; Lin et al., 2010). The BPS is similarly highly conserved, and is much more pyrimidine rich than that of U2-type introns. Conversely, no conserved PPT has been observed in U12-type introns. The 3'ss is marked only by the terminal nucleotides, which typically are YAG (where Y is a pyrimidine) in introns starting with GT, and YAC in introns with AT at the 5'ss, although many different dinucleotide combinations can be recognized as a 3'ss (Dietrich et al., 2001; Levine and Durbin, 2001; Hastings et al., 2005). However, the distance of a U12-type BPS from the 3'ss is typically much more restricted (10 20 bp) than for U2-introns, and it has been shown that the distance is a crucial factor for recognition of the 3'ss (Dietrich et al., 2001; Levine and Durbin, 2001; Zhu and Brendel, 2003; Dietrich et al., 2005) Conversion of intron types, and evolution of U12-type introns The existence of U12-type introns and/or U12-dependent splicing factors in almost all of the major eukaryotic lineages suggests that U12-type introns are of ancient origin, similarly to U2-type introns (Russell et al., 2006; Bartschat and Samuelsson, 2010). However, they are entirely absent from many eukaryotes whose relatives nonetheless have them, suggesting repeated loss of U12-dependent splicing during evolution. The loss seems to have been most prevalent among diverse eukaryotic microbes, such as algae, unicellular fungi and amoebozoa, but some animals, including some but not all nematodes, have also lost their U12-type introns (Russell et al., 2006; Bartschat and Samuelsson, 2010). Interestingly, the genome of the nematode Caenorhabditis elegans still contains vestigial U12-type introns that are now spliced by the U2-dependent spliceosome (Sheth et al., 2006), supporting the idea that its U12-type introns have been converted to U2-type relatively recently. As discussed in chapter , the loss of U12-type introns probably reflects the high selection pressures affecting intron density in general during the evolution of these organisms. How, then, do U12-type introns become lost? As the vestigial U12-type introns of C. elegans suggest, one likely pathway is the step-wise conversion of U12-type introns into U2-type introns. Natural examples as well as experimental approaches have also shown that simple point mutations in the 5'ss of a U12-type intron, especially in the GT-AG subtype, suffice to turn it into a U2-type 5'ss (Dietrich et al., 1997; Burge et al., 1998; Lin et al., 2010), as U2-7

16 Review of the literature type introns have much more degenerate splicing signals (Figure 1B). Furthermore, changes in the terminal dinucleotides of U12-type introns occur readily in nature (Sheth et al., 2006; Lin et al., 2010), thus enabling other U12-subtypes to become GT-AG introns. Sometimes the conversion can also occur by activation of a cryptic U2-type 5'ss near the original U12-type 5'ss (Bartschat and Samuelsson, 2010). It has been also noted that the pyrimidine-rich BPS of U12-type introns could efficiently function as a PPT for the newly created U2-type intron (Burge et al., 1998). Given that the chance of creating the highly stringent U12-type 5'ss from a U2-type 5'ss is low, conversion in the reverse direction is a highly unlikely event. In fact, only one instance of a novel U12-type intron has been observed (Lin et al., 2010). Thus, without natural selection acting in favour of maintaining U12-type introns, they are likely to be converted into U2-type introns during evolution. Such a unidirectional conversion from U12-type introns to U2-type has raised the hypothesis that U12-type introns would have been much more prevalent during the early history of life, and that they might in fact be ancestral to U2-type introns. A further indication of this could be the fact that U12-type introns appear to resemble group II introns more closely (for discussion, see Roy and Irimia, 2009), assuming that spliceosomal introns are indeed derived from group II introns (see chapter ). However, an analysis of amino acid distributions at intron insertion sites retained between human and Arabidopsis thaliana genomes suggested that most ancestral introns were already of the U2-type (Basu et al., 2008b). Therefore, the precise origins of both U2 and U12-type introns remain unresolved, although it is clear that both have been present in the early eukaryotic ancestors Significance of U12-type introns The apparent ease of loss of U12-type introns in some lineages and their low prevalence (0.35% of all introns in humans) might initially suggest that they are merely relics of the past that are fading away. However, the fact that they are retained in various highly divergent eukaryotes suggests they perform an essential function. It has been noted that U12-type introns are excised more slowly than U2-type introns in a wide variety of organisms from animals to plants (Patel et al., 2002; Lewandowska et al., 2004; Pessa et al., 2006), giving rise to the notion that their removal may be a rate-limiting step in gene regulation. U12-type introns generally occur only once in a gene (Burge et al., 1998; Levine and Durbin, 2001), suggesting that for such a rate-limiting function, only one U12-type introns is required. Furthermore, certain U12-type introns have retained their position in specific orthologous genes in distantly related organisms (Burge et al., 1998; Basu et al., 2008a). Together, these data suggest that the activity of U12-dependent splicing factors may have a conserved role in regulating a specific set of genes in response to external stimuli or in specific tissues or developmental stages. Indeed, U12-type intron-containing genes show a bias for being differentially expressed in bone marrow CD34 positive cells and B lymphoblast (Yeo et al., 2007). More importantly, while U12-type introns account for a very small fraction of Drosophila melanogaster introns (Schneider et al., 2004; Sheth et al., 2006), U12-dependent splicing is nonetheless essential for D. melanogaster development (Otake et al., 2002). U12-8

17 Review of the literature type introns are not randomly scattered in genomes, but are most often found in genes that can be loosely categorized as information processing genes (Burge et al., 1998; Yeo et al., 2007). These genes have functions in DNA replication and repair, transcription, pre-mrna processing, translation and signal transduction, suggesting that the essential function of U12- dependent spliceosome is linked to the regulation of these genes. However, alterations in U12- dependent splicing can also have wide effects on multiple downstream processes, as shown by the disruption metabolic functions in flies with defective U12-dependent splicing (Pessa et al., 2010). 1.4 Spliceosome components Components of the U2-dependent spliceosome The core components of the major spliceosome include five small nuclear ribonucleoprotein particles (snrnps), each of which contains, and is named after, one of the small nuclear RNAs (snrnas): U1, U2, U4, U5 and U6 (Figure 2). Each snrna, except U6, contains a binding site for Sm proteins, which together form a ring-like core structure for the snrnps, while U6 has a similar structure composed of Sm-like (Lsm) proteins (reviewed by Patel and Bellini, 2008). The snrnps also contain a varying number of specific protein factors (reviewed by Valadkhan and Jaladat, 2010): U1-specific proteins U1A and U1-70K bind directly to U1 snrna, while the U1C protein associates through protein-protein interactions (Query et al., 1989; Scherly et al., 1989; Nelissen et al., 1991; Pomeranz Krummel et al., 2009). In addition to the Sm proteins, the core of the U2 snrnp contains the proteins U2A' and U2B'', and is complemented by two multiprotein complexes, SF3a and SF3b (Boelens et al., 1990; Brosi et al., 1993; Krämer et al., 1999). U4 and U6 snrnas associate with one another through basepairs formed between highly conserved sequence elements (Bringmann et al., 1984; Hashimoto and Steitz, 1984), and together with specific proteins form the U4/U6 di-snrnp (Nottrott et al., 2002). U5 snrnp associates with U4/U6 through protein-protein interactions (Cheng and Abelson, 1987; Konarska and Sharp, 1988; Black and Pinto, 1989), forming the U4/U6-U5 tri-snrnp together with additional tri-snrnp-specific protein factors (Behrens and Lührmann, 1991; Liu et al., 2006 and references therein). U5 is the largest of the spliceosomal snrnps (Black and Pinto, 1989), and contains several large proteins that make up a significant part of the catalytically active spliceosome, most notably the multifunctional protein Prp8 (reviewed by Valadkhan and Jaladat, 2010). In addition to the snrnps, a number of non-snrnp proteins are also found among the core components of the U2-dependent spliceosome, including splicing factor 1 (SF1) and U2 auxiliary factor (U2AF), which have important functions for recognizing the BPS, PPT and 3'ss (See chapter 1.5.2). To facilitate the conformational changes occurring during assembly, the spliceosome employs a number of peptidyl-prolyl cis/trans isomerases, helicases, kinases and other protein and RNA-modifying enzymes, some of which are snrnp components while others are non-snrnp proteins (reviewed by Staley and Guthrie, 1998; Smith et al., 2008). The spliceosome also interacts with a large number of splicing factors that can suppress or 9

18 Review of the literature Figure 2. Secondary structures of spliceosomal snrnas. The predicted secondary structures of the human spliceosomal snrnas are depicted here schematically. The binding sites for Sm proteins are boxed, the sequences interacting with the 5'ss or BPS are marked with a black line, and sequences involved in U2/U6 or U12/U6atac basepairing are highlighted in gray. Adapted with modifications from Pessa (2010). Structures are as published by Yu et al. (1999) for U1, U2, U4, U5 and U6, Tarn and Steitz (1997) for U11, Sikand and Shukla (2011) for U12, and Padgett and Shukla (2002) for U4atac and U6atac. 10

19 Review of the literature enhance splicing at a given site, such as heterogeneous nuclear ribonucleoproteins (hnrnps) and serine-arginine-rich (SR) proteins (see chapter 1.6.4). Many of the protein factors associate with the spliceosome during specific stages of assembly, and can also be specific to certain introns or tissues only, making the spliceosome a very dynamic and variable molecular machine. It is therefore difficult to pinpoint the exact composition of the spliceosome, and estimates for the number of protein components vary between 150 and 300 (reviewed by Jurica and Moore, 2003; Valadkhan and Jaladat, 2010) Components of the U12-dependent spliceosome The U12-dependent spliceosome is similarly composed of five snrnps, U11, U12, U4atac, U5 and U6atac. U5 is therefore a common component of both spliceosomes, and U11, U12, U4atac and U6atac are structural and functional counterparts of U1, U2, U4 and U6, respectively (Hall and Padgett, 1996; Tarn and Steitz, 1996a, b; Kolossova and Padgett, 1997; Yu and Steitz, 1997; Incorvaia and Padgett, 1998). The U12-type snrnas do not resemble their U2-type counterparts in sequence: U11 and U12 are entirely different from U1 and U2 (Montzka and Steitz, 1988), and U4atac and U6atac share limited (ca. 40%) sequence similarity with U4 and U6, respectively (Tarn and Steitz, 1996a). However, the secondary structures of the analogous snrnas are highly similar, highlighting the similarity of their functions (Figure 2). The low abundance of minor introns is also reflected in the numbers of the minor snrnps, which are approximately 100 times less abundant in human than the major snrnps (Montzka and Steitz, 1988; Tarn and Steitz, 1996a). This has hindered the comprehensive analysis of their protein components, and consequently the composition of the minor snrnps and spliceosomal complexes are known in less detail than their major counterparts (see e.g. Schneider et al., 2002). However, most of the protein components of the two spliceosomes seem to be shared. The minor snrnas are complexed with Sm or Lsm proteins similarly to the major snrnas (Montzka and Steitz, 1988; Tarn and Steitz, 1996a; Will et al., 1999; Schneider et al., 2002). The minor U4atac/U6atac-U5 tri-snrnp also appears to contain most if not all of the protein components of the U4/U6-U5 tri-snrnp (Luo et al., 1999; Schneider et al., 2002). In contrast to U1 and U2, however, U11 and U12 snrnps interact with one another already before spliceosome assembly, and are mainly present as preformed U11/U12 disnrnps (Wassarman and Steitz, 1992; see chapter 1.5.4). The protein composition of the U11/U12 di-snrnp is also different from that of the U1 and U2 snrnps: U11/U12 entirely lacks U1-specific proteins and the U2-specific SF3a protein complex (Will et al., 1999). However, it does contain the SF3b complex and three other proteins present in the major spliceosome (YB-1, hprp43p and Urp), and additionally seven proteins (20K, 25K, 31K, 35K, 48K, 59K and 65K) specific to the U12-dependent spliceosome (Will et al., 1999; 2001; 2004). Four of these (25K, 35K, 48K, 59K) are also found in free U11 snrnps, while three appear to associate only with the di-snrnp (Will et al., 2004). Although their functions remain undeciphered for the most part, their association with the U11/U12 di-snrnp suggests they are involved with 5'ss and BPS recognition and in interactions bridging these two sites (see chapter 1.5.5). 11

20 Review of the literature 1.5 Spliceosome assembly and catalysis Splicing catalysis Both spliceosomes catalyze the same, two-step transesterification reaction that results in the formation of an excised intron lariat and ligation of the two exons (Figure 3; reviewed by Will and Lührmann, 2011). In the first step, the 2' hydroxyl group of the branch point adenosine performs a nucleophilic attack on the phosphate group at the 5'ss, resulting in formation of the lariat structure and a free 3' OH group at the end of the upstream exon. This 3' OH then attacks the phosphate at the 3'ss, resulting in the ligation of the two exons, and release of the intron lariat. Due to its nature as a transesterification reaction, splicing in itself is energetically neutral. However, ATP is consumed in many energy-requiring steps during the assembly of the spliceosome to ensure the specificity and unidirectionality of the reaction. Figure 3. Catalytic mechanism of splicing. Splicing occurs through a twostep transesterification reaction, depicted here schematically. Intron stucture is depicted as in Figure 1A, and additionally the reactive phosphates (P) and hydroxyl groups (OH) are indicated. In the first step of splicing, the 2' hydroxyl group of the BP adenosine attacks the 5'ss phosphate, resulting in the formation of a branched intron lariat structure, and a free 3' hydroxyl in exon 1. In the second step, this hydroxyl attacks the phosphate at the 3'ss, resulting in the ligation of the exons and release of the intron lariat Assembly of the U2-dependent spliceosome Unlike another large ribonucleoprotein machinery, the ribosome, the components of the spliceosome must undergo a series of complicated rearrangements to become catalytically activated (Figure 4). While pre-formed higher order structures of snrnps have been observed in some cases (Konarska and Sharp, 1988; Stevens et al., 2002; Malca et al., 2003), the interactions of snrnps with each other and with the pre-mrna are generally established in a 12

21 Review of the literature step-wise manner (Tardiff and Rosbash, 2006; Huranová et al., 2010). The snrnps are loosely associated with the pre-mrna already at the earliest stages of spliceosome assembly, but their interactions with the pre-mrna are tightly controlled in order to avoid premature and unspecific activation of the spliceosome (reviewed by Smith et al., 2008; Wahl et al., 2009). The reactive sites in the pre-mrna are recognized multiple times by RNA or protein factors, a process often referred to as proofreading. The recognition events generally depend on interactions that are weak on their own, but are stabilized by the combined effects of multiple factors. Such intrinsically weak interactions also allow the exchange of binding partners, which is crucial for remodeling of the spliceosome at different stages of its assembly. Thus, the spliceosome is a highly dynamic molecular machine that undergoes dramatic changes both in its composition and its conformation during its maturation. The association of snrnps and other spliceosome components with the pre-mrna during spliceosome assembly are often described in terms of complexes (E, A, B, B act, B* and C; Figure 4) that can be separated by biochemical methods (reviewed by Will and Lührmann, 2011). The initial step in intron recognition is the formation of the commitment (or E) complex, during which U1 snrnp binds to the 5'ss sequence. U1 snrna base-pairs to the 5'ss through a complementary sequence at its 5' end (Zhuang and Weiner, 1986), and this interaction is stabilized by protein-rna interactions, most significantly by the U1C protein, which stabilizes the 5'ss/U1 helix (Heinrichs et al., 1990; Pomeranz Krummel et al., 2009). During this stage, the BPS is recognized by the protein factor SF1 (Berglund et al., 1997), which defines the catalytically functional adenosine such that it bulges out of the structure (Liu et al., 2001), making it available for the subsequent nucleophilic attack in the later stages of spliceosome maturation. At the same time, the PPT is recognized by the 65 kda subunit of U2AF (Zamore and Green, 1989), while its 35 kda subunit interacts with the 3'ss (Guth et al., 1999; Wu et al., 1999; Zorio and Blumenthal, 1999). Although the U2 snrnp is already present in the E complex and in proximity to U1 (Dönmez et al., 2007), it becomes stably associated with the BPS in the ATP-dependent A complex, also known as the prespliceosome (Bindereif and Green, 1987; Liao et al., 1992). During the formation of the prespliceosome, U2AF recruits U2 snrnp to the BPS through interactions with the SF3b complex, displacing SF1 (Ruskin et al., 1988; Valcárcel et al., 1996; Gozani et al., 1998; Rutz and Séraphin, 1999; Spadaccini et al., 2006; see also Figure 4). Thus, in a proofreading event typical of the spliceosome, the task of defining the BP is transferred from SF1 to U2 snrnp. U2 snrna associates with the BPS through base-pairing interactions (Wu and Manley, 1989; Zhuang and Weiner, 1989), which exclude the reactive adenosine (Query et al., 1994), therefore retaining the bulged conformation. The specific recognition of the BP is additionally mediated by proteins of the SF3b complex, which together with the SF3a complex is also involved in stabilizing U2 binding to the BPS and to U2AF (Gozani et al., 1996; 1998; Spadaccini et al., 2006). 13

22 Review of the literature Figure 4. Spliceosome assembly. The interactions of the spliceosomal snrnps and some selected non-snrnp protein complexes at various stages of spliceosome assembly are depicted schematically for both the U2-dependent and U12- dependent spliceosomes, as indicated above the panel (see chapter 1.5 for details). Complexes E, A, B* and C are indicated in the middle.the Prp19/CDC5 complex is indicated by "19C". The association and dissociation of certain protein complexes is not known in detail for the U12-dependent spliceosome, and such events are marked with question marks. 14

23 Review of the literature The U4/U6-U5 tri-snrnp is also present already in the E complex, with the U5-specific protein Prp8 in direct contact with the 5'ss (Maroney et al., 2000). However, compositional and conformational changes during B complex formation lead to the stable association of U4/U6-U5 tri-snrnp (Konarska and Sharp, 1987; Lamond et al., 1988), followed by a number of significant rearrangements in RNA-RNA interactions to yield the pre-catalytic B act complex: In another significant proofreading event, U1 is displaced from the 5'ss, which instead becomes base-paired to U6 (Kandels-Lewis and Séraphin, 1993). The helices between U4 and U6 snrnas unwind and U4 snrnp is released from the spliceosome (Konarska and Sharp, 1987; Lamond et al., 1988). U6 base-pairs to U2, forming RNA structures necessary for splicing catalysis (Wu and Manley, 1991; Madhani and Guthrie, 1992). U5 snrna is not involved in sequence-specific base-pairing interactions, but seems to stabilize spliceosome structures by binding to both exons via its U-rich loop structure (Newman and Norman, 1992; Sontheimer and Steitz, 1993). Interactions between the snrnas are not the only ones to be remodeled at this stage. Many of the snrnp proteins dissociate, including all those of U1, U4 and U6, as well as a number of non-snrnp proteins (Bessonov et al., 2008; Fabrizio et al., 2009; Agafonov et al., 2011). The spliceosome is also joined by a number of factors, including the Prp19/CDC5 complex, which is involved in rearranging many protein-protein and protein-rna interactions, particularly those of U5 and U6 (Makarov et al., 2002; Chan et al., 2003; Makarova et al., 2004). These changes eventually yield the activated spliceosome, or complex B* (Figure 4). The U5- specific Prp8 protein likely functions as a crucial platform for the catalytic core, as it forms a network of interactions with the catalytically active sites during the maturation of the spliceosome, eventually interacting with both splice sites and the BPS, as well as U5 and U6 snrnas (reviewed by Grainger and Beggs, 2005). After the first catalytic step, the SF3a and SF3b complexes of U2 snrnp dissociate, and the spliceosome is joined by further factors, including specific helicases and peptidyl-prolyl cis/trans isomerases (Bessonov et al., 2008; Fabrizio et al., 2009; Agafonov et al., 2011). These likely facilitate the conformational changes required for formation of complex C, which catalyzes the second catalytic step Is the spliceosome a ribozyme? The similarities between the spliceosome and self-splicing group II introns suggest that it may also be an RNA enzyme (see chapter ). Multiple lines of evidence suggest that U6 snrna has a crucial role in the catalytic center. All U6-specific proteins, as well as the U4 snrnp, are released during the maturation of the catalytic spliceosome (Bessonov et al., 2008; Agafonov et al., 2011), suggesting that their function is merely to escort U6 snrna into the catalytic core. Moreover, the binding of Mg 2+ ions by U6 has been implicated in catalysis (Yean et al., 2000; Lee et al., 2010), and the first step of splicing takes place in the vicinity of an invariant sequence of U6 (reviewed by Valadkhan, 2010). Most strikingly, U2 and U6 snrnas can catalyze a two-step splicing reaction between short oligonucleotides in the absence of proteins (Valadkhan et al., 2009). However, it is clear that proteins play significant roles for the specificity, efficiency and fidelity of the spliceosome. Recent evidence also suggests that protein components of the spliceosome, particularly Prp8 and its RNase H-like 15

24 Review of the literature domain, might be involved in the catalysis itself (reviewed by Abelson, 2008). It is likely that both proteins and RNA contribute to the catalytic activity, but the question remains unresolved for the time being Assembly of the U12-dependent spliceosome The overall assembly pathways in the two spliceosomes are similar (reviewed by Patel and Steitz, 2003). However, the initial recognition of U12-type introns differs from that of U2-type introns (Figure 4). The preformed U11-U12 di-snrnps bind the intron as a unit, and the 5'ss and BPS are recognized in a co-operative manner, although U11/5'ss base-pairing can still be detected prior to stable base-pair formation between U12 and the BPS (Frilander and Steitz, 1999). Consequently, the first observed complex for U12-type introns in native gels is the ATP-dependent A complex (Tarn and Steitz, 1996b; Frilander and Steitz, 1999; Figure 4). U12-type introns do not have PPTs, and U2AF is not required for the recognition of U12-type introns (Shen and Green, 2007). However, a U2AF35-related protein, Urp, is required for A complex formation and 3'ss recognition (Shen et al., 2010). Interestingly, Urp is also required for U2-dependent splicing, but only after the first catalytic step, when it apparently displaces U2AF from the PPT and 3'ss (Shen et al., 2010), suggesting that the U2-type 3'ss is proofread several times by protein factors. In contrast, the lack of PPT, more conserved recognition sequences and more restricted BPS-3'ss distance (Figure 1) suggest that the recognition of U12-type introns relies more on RNA-RNA interactions, and protein-mediated proofreading is less important (Brock et al., 2008). As stated above (chapter 1.4.2), due to the low abundance of U12-type factors, our understanding of the specific interactions during minor spliceosome assembly is less complete than for the major spliceosome. However, despite the differences in intron recognition, it seems that the steps leading to catalytic core formation are similar to those of the major spliceosome (Figure 4). The U12/BPS duplex is highly analogous to the U2/BPS duplex, causing the bulging of the reactive adenosine (Tarn and Steitz, 1996b). The SF3b complex associates with U12 as well as U2 snrnps, and helps to define the BPS in both spliceosomes (Gozani et al., 1996; Will et al., 2001; 2004). Upon formation of the B complex, the U4atac/U6atac/U5 tri-snrnp joins the forming spliceosome, leading to the displacement of U11 from the 5'ss by U6atac, which also base-pairs to U12, with the concomitant release of U4atac (Tarn and Steitz, 1996a; Yu and Steitz, 1997; Incorvaia and Padgett, 1998). The modifications taking place in the RNA-RNA interaction network are similar to those of the major spliceosome, but the order of events appears to be somewhat more flexible in the minor spliceosome (Frilander and Steitz, 2001). The active spliceosome then catalyses the two-step transesterification reaction, resulting in ligation of the exons and release of the intron lariat (Tarn and Steitz, 1996b; see Figures 3 and 4). Due to the identical chemistries of the splicing reactions, the catalytic cores of the two spliceosomes are likely to be similar. In support of this, the functional domains of U6 and U6atac snrnas are highly similar, and are functional in splicing when replaced with one another (Shukla and Padgett, 2001). 16

25 Review of the literature Comparison of initial intron recognition in the two spliceosomes The absence of U2AF and SF1, as well as the lack of an E complex during U12-dependent prespliceosome assembly, have led to the suggestion that recognition of the minor introns relies more on snrna-rna interactions, and requires less proofreading by protein factors than recognition of major introns (chapter 1.5.4; for discussion, see also Patel and Steitz, 2003). This may also be the consequence, at least partially, of the co-operative recognition of the minor 5'ss and BPS by the preformed U11/U12 di-snrnp (Frilander and Steitz, 1999), which imposes more stringent requirements on sequence recognition. However, snrnas are not alone in recognizing the splice sites in the minor prespliceome. As mentioned above (chapter 1.5.4), proteins factors of the SF3b complex are involved in recognition of the U12- type BPS, although the SF3a complex present in the U2-type prespliceosome is absent (Will et al., 1999; 2004). Interestingly, in contrast to major prespliceosome, the first three nucleotides of the U12-type introns (RUA; see Figure 1B) are not recognized by snrnas at all, suggesting that specific protein factors are involved (see chapter 4.1). In the major spliceosome, the 5'ss/U1 helix spans the exon-intron junction, and is stabilized by interactions with the U1C protein. It has been noted that the U11/U12-20K protein has sequence similarity with U1C (Will et al., 2004), but no interaction with the 5'ss and the 20K protein has been observed. U1/5'ss interactions are also stabilized by SR proteins, particularly SRSF1, which binds directly to the U1-70K protein (see chapter ). The U11-35K protein shares similarity with U1-70K (Will et al., 1999), and both have been observed to interact with the homologs of SRSF1 in Arabidopsis thaliana (Lorković et al., 2004), suggesting that 5'ss recognition can be enhanced by similar mechanisms in both spliceosomes. SR proteins have been shown to interact with both the 5'ss and BPS in both spliceosomes (Shen and Green, 2006, 2007), and may thus contribute to interactions that bring these two sites together. However, in the minor prespliceosome, the internal components of the U11/U12 di-snrnp are obviously involved in bridging the 5'ss and the BPS (Benecke et al., 2005; I). Interestingly, the catalytically active 5' end of U12 is also brought close to the 5'ss during prespliceosome formation, suggesting that major rearrangements in the conformation of U12 are not necessary at later stages of spliceosome assembly (Frilander and Meng, 2005). Similarly, the U2-type 5'ss is in proximity to both the BPS and U2 already in the E complex, before U2 base-pairs to the BPS (Kent and MacMillan, 2002; Dönmez et al., 2007). The 5' region of U2 snrna has been shown to be close to the stem-loop SL3 of U1 snrna, suggesting it might function as a binding platform for U2 snrnp (Dönmez et al., 2007; Weber et al., 2010). However, a number of non-snrnp proteins, including Prp5, have also been implicated in bridging U1 and U2 snrnps (Xu et al., 2004; Shao et al., 2011), indicating that internal snrnp components are not sufficient for intron bridging in the major spliceosome. It is likely that these differences in intron recognition are used by spliceosome-specific regulatory mechanisms to control their activity. 17

26 Review of the literature 1.6 Splice site definition and alternative splicing Initial splice site definition over exons In many multicellular eukaryotes, especially vertebrates, primary transcripts are mainly composed of intronic sequences, and some of the introns may be hundreds of kilobases long. This poses the splicing machinery with a difficult problem of correctly pairing splice sites that are separated by huge distances. On the other hand, vertebrate exons are generally rather short, suggesting that the initial splice site pairing in organisms with long introns in fact takes place over exons (Figure 5; reviewed by Berget, 1995). Indeed, 3'ss-recognizing factors of the upstream intron can interact with the 5'ss-recognizing factors of the downstream introns (Robberson et al., 1990; Hoffman and Grabowski, 1992), and long exons flanked by long introns are recognized poorly (Sterner et al., 1996; Fox-Walsh et al., 2005). Such exon definition interactions are later replaced with interactions that pair the splicing factors within the same intron. The final splice site pairing generally takes place in the A complex (Lim and Hertel, 2004; see also Figure 4), but apparently the process is flexible, as direct conversion of exon-defined complexes into B complexes has also been observed (Schneider et al., 2010). Exon definition appears to be particularly important in mammals. The most common form of aberrant splicing in mammals is exon skipping (Nakai and Sakamoto, 1994), which is consistent with pairing of splicing factors located at the upstream and downstream exon, when defining the middle exon fails. The importance of exon definition in mammals is also apparent by the fact that mutations weakening one splice site flanking an exon are often compensated by mutations strengthening the other (Xiao et al., 2007). In contrast, in organisms with shorter introns, such as plants, fungi and many invertebrates, compensatory mutations generally occur within the same intron (Xiao et al., 2007), suggesting that shorter introns are defined directly, as also experimentally observed for mammalian introns by Sterner et al. (1996) and Fox- Walsh et al. (2005) Splicing enhancers and silencers Due to their short length and, especially in U2-type introns, degenerate nature, the core splicing signals within introns are not sufficient for unambiguous definition of splice sites (Burge et al., 1999; Lim and Burge, 2001). Furthermore, true U2-type 5'ss signals are outnumbered many-fold by similar, non-functional splice sites (pseudo splice sites), and are often found in positions where they could define putative exons (pseudoexons) with upstream 3'ss-like sequences. Thus, the correct activation of splice sites requires additional information, which is provided by enhancer or silencer sequences in the vicinity of the splice sites (Figure 5). They are categorized based on their location into exonic splicing enhancers (ESEs) or silencers (ESSs) or intronic splicing enhancers (ISEs) or silencers (ISSs). Although each of these splicing regulatory element (SRE) subtypes has some typical characteristics, they are composed of short and variable sequences. As they can function in combinations, and either in a co-operative or antagonistic manner, their effects on splicing are highly dependent on context, and are difficult to predict based on sequence information alone (for review, see Wang and Burge, 2008). However, large-scale bioinformatic studies have provided 18

27 Review of the literature information about SRE distributions in true exons and pseudoexons, and the effect of short sequence motifs on splicing in vitro and in vivo has been tested experimentally by selective molecular evolution methods (reviewed by Chasin, 2007). Most promisingly, some progress has recently been made in this field by using computational methods to analyze and predict alternative splicing patterns in different tissues (Barash et al., 2010). Among other things, this study revealed that alternative splicing is affected by intronic sequence elements located much further away ( nt) from splice sites than previously thought. While it has been established that the more conserved U12-type introns do contain enough information to be recognized correctly in a given transcript, their presence in only a handful of transcripts means that specifying their unique location within all transcripts requires additional information (Burge et al., 1999). Accordingly, they also respond to regulation by SREs (Wu and Krainer, 1998; Hastings and Krainer, 2001; Lewandowska et al., 2004), and U12- dependent splicing is enhanced by exon definition interactions with U2-type factors in neighbouring introns, and vice versa (Wu and Krainer, 1996; Lewandowska et al., 2004; II). Figure 5. Splicing regulatory elements and splice site definition. Splicing is regulated by trans-acting factors that bind to enhancer and silencer elements. These are also involved in bringing together spliceosomal components bound at the 5'ss and 3'ss, either through exon definition or intron definition interactions. Exons and the regulatory elements (ESEs, ESSs) within them are depicted as boxes, and introns and their regulatory elements (ISEs, ISSs) by lines. Interactions between splicing factors are depicted schematically, with arrows representing activating interactions, and blocked lines representing inhibitory interactions. See chapter 1.6 for details Alternative splicing increases proteome complexity Regulation of splicing by SREs is important not only for distinguishing true exons from pseudoexons, but also for producing alternatively spliced mrnas. Alternative splicing is the most significant mechanism for increasing proteome diversity, accounting for most of the estimated 5-fold excess of proteins over protein-coding genes in humans (reviewed by Nilsen and Graveley, 2010). In fact, it is utilized by almost all protein-coding genes, as recent studies indicate that % of human multi-exon pre-mrnas are spliced to produce at least two abundant isoforms (Pan et al., 2008; Wang et al., 2008). A typical case of alternative splicing in mammals involves the inclusion or exclusion of an alternative cassette exon (Figure 6C), reflecting the importance of exon definition interactions in these organisms (see chapter 1.6.1). Other, slightly less common modes of alternative splicing include use of alternative 5'ss or 3'ss, mutually exclusive exons or intron inclusion (Figure 6A, 6B, 6D and 6E, respectively). Thus, the majority of alternative splice site choices seem to take place between competing splice sites during formation of the commitment complex, either through exon definition or 19